[campaign] ReScript → AffineScript estate migration (UMBRELLA)

[hyperpolymath]

Campaign overview

Estate-wide migration of ReScript (.res, .resi) to AffineScript. Parallel to the TS→AffineScript campaign (#239) but ~14× the scale.

State: planning. NO step-issues filed yet (per operator directive 2026-05-28). All planning captured here; step-issues to be filed when work begins.

Scope safety: only hyperpolymath/ repos that are non-fork. Top 16 backlog repos verified owner=hyperpolymath, not-fork (2026-05-28). isers and affinescript-stdlib-pr returned UNRESOLVED (likely renamed or local-only — verify before touching).

Inventory snapshot 2026-05-28

• 3,996 total .res / .resi files across ~80 repos (post-excludes: node_modules, dist, build, target, _build, _opam, rescript/ upstream, servers/, repos-monorepo/, linguist/, hyperpolymath-archive/, deps/, _wt-*, .deno, vendor)
• ~14× the TypeScript backlog (~285 files / 38 repos in [campaign] TypeScript → AffineScript estate migration (UMBRELLA) #239)
• Largest single target: idaptik (1,235 files) — vs TS's 52
• Top-3: idaptik 1235 / panll 726 / developer-ecosystem 460 = 60% of backlog

4-layer architecture (mirrors TS campaign #239)

LAYER 1 — POLICY & ENFORCEMENT  PARTIAL
  hypatia :rescript_detected (*.res) + :rescript_interface_detected (*.resi) EXIST
  but lack path_allow_prefixes (same gap TS had before hypatia#375).
  NEEDS: path_allow_prefixes for carve-outs analogous to TS 9-class
  canon: bsconfig.json tooling, *.config.res, upstream forks, vendored
  /deps/, /vscode/, archived, and any RS-specific (e.g. lib/js/ build
  output if not already excluded).
  standards docs: LANGUAGE-POLICY.adoc ReScript row, .claude/CLAUDE.md
  exemption table, language-policy.yml CI gate — all need updates
  paralleling TS-layer-1 PR set.

LAYER 2 — STDLIB BINDING CAPACITY  PARTIAL (shared with TS)
  AS stdlib coverage drives migration feasibility. Most RS that we own
  is application code, often consuming Belt/Js/Promise/RescriptCore —
  AS needs equivalents for each RS-side surface used.
  Likely gaps to enumerate from real ports: Belt collections, Js.Promise,
  RescriptCore HTTP, ReScript-React (if any), Bs.* legacy bindings.

LAYER 3 — PHYSICAL FILE PORTS  (~3996 files)
  Realistic estimate: 200-400 PRs over many months.
  Cannot use "smallest-first" naively — idaptik and panll dominate.
  Better strategy: per-repo or per-subsystem batches.

LAYER 4 — PARALLEL DEPENDENCY: AS bindings top-50 roadmap (#245 cross-ref)

Carve-out classes (to be encoded in hypatia rule)

Adapted from TS 9-class canon:

#	Class	Pattern	Notes
1	declaration / interface	*.resi separately tracked	Often paired with .res; carve-out applies per-pair
2	tooling	bsconfig.json, *.config.res	Build orchestration
3	upstream forks	rescript/, rescript-ecosystem/ (verify ownership)	Not estate-authored
4	bootstrap shims	TBD per-discovery	RS used to bootstrap AS toolchain?
5	archived	hyperpolymath-archive/**	Dormant
6	vendored deps	**/deps/**, **/node_modules/**	Same as TS
7	editor extensions	**/vscode/**	Same as TS (also editors/vscode/extensions/vscode/clients/vscode)
8	compiled output	**/lib/js/**, **/lib/es6/**, **/lib/bs/**	RS compiles to these — exclude already excluded by dist/build but lib/js/ may not be

Recommended step ordering (DEFERRED — file as step-issues when work begins)

1. STEP 1: Layer-1 PR set — hypatia path_allow_prefixes (analog of hypatia#375+Hypatia reusable SARIF converter can re-upload code-scanning mirror findings #378) + standards mirror (analog of standards#235+docs+ci: mirror hypatia#378 TS carve-out extension (linguist + /deps/ + /vscode/) #238+docs(ROADMAP): add TS→AffineScript estate migration campaign section #247)
2. STEP 2: Inventory verification — re-run with refined excludes, generate per-repo backlog table
3. STEP 3: Pick seed-port repo — smallest pure-logic with no Belt/Js dep
4. STEP 4: Stdlib gap-fill from step-3 surface
5. STEP 5: Smallest-batch tail-end repos (≤20 file repos × ~50 of them)
6. STEP 6: Mid-tier repos (20-100 file × ~15 of them)
7. STEP 7: Large repos (100-500 files × ~6 of them)
8. STEP 8: idaptik DLC + panll batched ports (1235+726 files = 50% of total)
9. STEP 9 (parallel-runnable post-bindings, NOT linearly blocked by STEPS 5-8): Plugin-host bound repos (zotero-tools, wordpress-tools — gated by AS host bindings same as TS step 6/7 / [campaign #239] STEP 6 — PLUGIN-HOST BINDINGS: cross-ref AS top-50 roadmap #245+[campaign #239] STEP 7 — PLUGIN REPOS: ~70 files in 5 repos (post step 6) #246). Can start any time after Layer-4 plugin-host bindings ship.

Cross-references

• TS umbrella: [campaign] TypeScript → AffineScript estate migration (UMBRELLA) #239 (same playbook, smaller scale)
• AS bindings parallel: project_affinescript_bindings_top50_roadmap.md
• ReScript migration assistant: [Repo] Migration assistant tool: parser + metaparser for .res→.affine skeletons affinescript#57 (existing tooling)
• Memory tracker: TBD (file when STEP 1 begins)
• Project board: https://github.com/users/hyperpolymath/projects/35

Scale summary

Metric	TS (#239)	ReScript (this)
Files	~285	3,996 (~14×)
Repos	38	~80 (~2×)
Largest single	idaptik 52	idaptik 1,235
PR estimate	50-80	200-400
Session estimate	8-15	multi-month